8  Relabeling Qualtrics Variables

Qualtrics uses some random values like -99 to encode for things like skipped questions, “not sure” categories, or “not applicable” answers. Currently one needs to a dictionary to look up every variable to understand what each value means.

In addition, some response categories are confusing because they equate to a missing response. For example, from an analytical perspective an answer of “not sure” and skipping the question completely both equate to a missing value when the response categories are yes or no.

The memisc package was designed for working with survey data. It allows you to label response categories in the data and also designate different types of missingness (e.g. Unsure, N/A, and -99 can all be coded as missing).

Several factor and boolean variables in this survey data set have inconsistent coding. They are recoded in this section and the decision criteria are documented for inspection and review.

9 Yes/No Questions

Original Value	Description	Recode Value	Code as Missing ?
Yes	Yes	1	No
No	No	0	No
Unsure	Unsure	97	Yes
N/A	Not Applicable	98	Yes
-99	Incomplete	99	Yes
NA	Unanswered	NA	Yes

# APPLY TO COLUMNS K:
bool_qns <- 
  c( program_change_qns_bool, 
     fundraise_qns_bool, 
     finance_chng_qns_bool, 
     leadership_chng_qns_bool )

COLUMNS <-  bool_qns

# VALUES THAT NEED RECODING
RULES <- c(
  
           "  Yes    =>>    1    ", 
           "  No     =>>    0    ", 
           "  Unsure =>>    97   ", 
           "  N/A    =>>    98   ", 
           "  -99    =>>    99   "   )

rules <- parse_rules( RULES )          
pattern <- rules[[ "pattern" ]]
replace <- rules[[ "replace" ]]


# MEMISC LABELS AND MISSING VALUE CODES
values  <- c( 0, 1, 97, 98, 99 )
labels  <- c( "No", "Yes", "Unsure", "Not Applicable", "N/A" )
missing <- c( 97, 98, 99 )

# RECODE VARIABLES 
survey_df <- 
  survey_df %>% 
  recode_columns( k=COLUMNS, pattern, replace, values, labels, missing )

Example:

PrgSrvc_IncrNum — ‘TYPE: boolean’

“Indicates an increase in number of programs or services”

Storage mode:	character
Measurement:	nominal
Missing values:	97, 98, 99

Values and labels			N	Valid			Total
0		‘No’	192	38	.	7	38	.	2
1		‘Yes’	304	61	.	3	60	.	6
98	M	‘Not Applicable’	3				0	.	6
99	M	‘N/A’	1				0	.	2
NA	M		2				0	.	4

QUESTION TXT:

In the last year (between January 2022-December 2022), did your organization make any of the following changes to your Programs, as compared to 2021? - Increased the number of programs or services

Original Value	Description	Recode Value	Code as Missing ?
Yes	Yes	1	No
No	No	2	No
Unsure	Unsure	3	Yes
N/A	Not Applicable	98	Yes
-99	Incomplete	99	Yes
NA	Unanswered	NA	Yes

# APPLY TO COLUMNS K:

COLUMNS <- cares_qns_bool

# VALUES THAT NEED RECODING
RULES <- c(  "  Yes    =>>  1    ", 
             "  No     =>>  2    ", 
             "  Unsure =>>  3    ", 
             "  N/A    =>>  98   ", 
             "  -99    =>>  99   "   )

rules <- parse_rules( RULES )          
pattern <- rules[[ "pattern" ]]
replace <- rules[[ "replace" ]]


# MEMISC LABELS AND MISSING VALUE CODES
values  <- c( 1, 2, 3, 98, 99 )
labels  <- c( "Yes", "No", "Unsure", "Not Applicable", "N/A" )
missing <- c( 3, 98, 99 )

# RECODE VARIABLES 
survey_df <- 
  survey_df %>% 
  recode_columns( k=COLUMNS, pattern, replace, values, labels, missing )

Example:

CARES_Rcv — ‘TYPE: boolean’

“Indicates whether organization receive government funding from CARES act”

Storage mode:	character
Measurement:	nominal
Missing values:	3, 98, 99

Values and labels			N	Valid			Total
1		‘Yes’	172	41	.	1	34	.	3
2		‘No’	247	58	.	9	49	.	2
3	M	‘Unsure’	16				3	.	2
NA	M		67				13	.	3

QUESTION TXT:

In the last year (2022, or your last fiscal year), did your organization receive any government funding to help your organization respond to the COVID-19 pandemic (including funding from federal, state, or local government programs)?

10 Single Checkboxes

These questions are presented as a checkbox to the respondent. They indicate an affirmative answer to the question.

10.1 Seek or Receive Fundraising Questions

These checkboxes are ticked by the respondent to indicate if they have sought or received funding from a specific source.

Original Value	Description	Recode Label	Recode Value	Code as Missing?
1	Checkbox Checked	Yes	1	No
NA	Checkbox Unchecked	No	0	No
All NA	Unanswered	NA	NA	Yes

ALL NA refers to no responses on all questions in this entire question group (i.e. the respondent skipped the section).

# APPLY TO COLUMNS K:
COLUMNS <-  fundraise_skrcv_qns_bool


# ALL RESPONSES NA IN GROUP:
# indication that the question
# was skipped by the respondent

row.id <- apply( survey_df[ COLUMNS ], 
                 MARGIN = 1, 
                 FUN = function(x) all(is.na(x)) ) 

# NA means "no"

na_to_zero <- function(x){
  x[ is.na(x) ] <- 0
  return(x)
}

survey_df[ COLUMNS ] <- 
  survey_df[ COLUMNS ] %>% 
  lapply( na_to_zero )

# if they do not respond "yes" 
# to at least one question then
# assume they skipped them all

zero_to_na <- function( x, rowid ){
  x[ rowid ] <- NA
  return(x)
}

survey_df[ COLUMNS ] <- 
  survey_df[ COLUMNS ] %>% 
  lapply( zero_to_na, row.id )



# VALUES THAT NEED RECODING

RULES <- c(    
  
  "    1      =>>    1    ", 
  "   NA      =>>    0    "    )


rules <- parse_rules( RULES )          
pattern <- rules[[ "pattern" ]]
replace <- rules[[ "replace" ]]

# MEMISC LABELS AND MISSING VALUE CODES 
values  <- c( "0", "1" )
labels  <- c( "No", "Yes" )
missing <- NA

# RECODE VARIABLES 
survey_df <- 
  survey_df %>% 
  recode_columns( k=COLUMNS, pattern, replace, values, labels, missing )

Example:

FndRaise_LocGvtGrnt_Seek — ‘TYPE: boolean’

“Indicates if nonprofit sought a local government grant”

Storage mode:	character
Measurement:	nominal
Missing values:	NA

Values and labels			N	Percent
0		‘No’	228	45	.	4
1		‘Yes’	274	54	.	6

QUESTION TXT:

In the last year (2022), did your organization seek or receive revenue from any of the followi… - Did you SEEK or APPLY for this funding? - Local government grants - (select all that apply)

10.2 Regulation Checkboxes

These questions are presented as a checkbox to the user to indicate that a question is not applicable. “Yes” here means “Yes, this question is not applicable”.

Original Value	Description	Recode Label	Recode Value	Code as Missing ?
1	Checkbox Checked	Yes	1	No
NA	Unanswered	No	0	No
22	Checkbox Checked	Yes	1	No
21	Unchecked Box	No	0	No

# APPLY TO COLUMNS K:

# exclude OTHER TXT question
other <- which( regulation_qns == "Regulations_Other_TEXT" )
COLUMNS <-  regulation_qns[ - other  ]


# "did not contact any agencies":
# ~ NONE CONTACTED

dnc.rows <- survey_df$Regulations == 21 

# all rows NA: 
# question skipped by respondent

na.rows <- is.na( survey_df$Regulations )


# NA means "NO"

na_to_zero <- function(x){
  x[ is.na(x) ] <- 0
  return(x)
}

survey_df[ COLUMNS ] <- 
  survey_df[ COLUMNS ] %>% 
  lapply( na_to_zero )


# if they responded "NONE CONTACTED" (21) 
# to the first question code
# the rest as "NONE CONTACTED"

zero_to_dnc <- function( x, rowid ){
  x[ rowid ] <- "NONE"
  return(x)
}

survey_df[ COLUMNS ] <- 
  survey_df[ COLUMNS ] %>% 
  lapply( zero_to_dnc, rowid=dnc.rows )

# if the first question was NA 
# code all as missing ("no response")

zero_to_noresp <- function( x, rowid ){
  x[ rowid ] <- "_NR_"
  return(x)
}

survey_df[ COLUMNS ] <- 
  survey_df[ COLUMNS ] %>% 
  lapply( zero_to_noresp, rowid=na.rows )


# VALUES THAT NEED RECODING

RULES <- c(    "   22    =>>     1                ",
               "   21    =>>     NONE             ",
               "   NA    =>>     _NA_             "    )


rules <- parse_rules( RULES )          
pattern <- rules[[ "pattern" ]]
replace <- rules[[ "replace" ]]


# MEMISC LABELS AND MISSING VALUE CODES 
values  <- c( "0", "1", "NONE", "_NR_", "_NA_" )
labels  <- c( "No", "Yes", "None Contacted", "No Response", "Missing" )
missing <- "_NA_"

# RECODE VARIABLES 
survey_df <- 
  survey_df %>% 
  recode_columns( k=COLUMNS, pattern, replace, values, labels, missing )

Example:

Regulations — ‘TYPE: boolean’

“Indicates whether the nonprofit contacted federal, state, or local regulatory agencies in the past 2 years”

Storage mode:	character
Measurement:	nominal
Missing values:	NA

Values and labels			N	Percent
NR		‘No Response’	63	12	.	5
0		‘No’	0	0	.	0
1		‘Yes’	156	31	.	1
NONE		‘None Contacted’	283	56	.	4

QUESTION TXT:

Has your organization contacted federal, state or local regulatory agencies (IRS, State Attorney Generals, Secretaries of State, or other state or local charities regulation officials) in the past 2 years?

Regulations_Local — ‘TYPE: boolean’

“Indicates whether the nonprofit has contacted local regulatory agencies in the past 2 years”

Storage mode:	character
Measurement:	nominal
Missing values:	NA

Values and labels			N	Percent
NR		‘No Response’	63	12	.	5
0		‘No’	60	12	.	0
1		‘Yes’	96	19	.	1
NONE		‘None Contacted’	283	56	.	4

QUESTION TXT:

Please indicate below the types of agencies (check all that apply): - Local

10.3 Staffing Plans

These questions are presented as a checkbox to the user to indicate that a question is not applicable. “Yes” here means “Yes, this question is not applicable”.

Original Value	Description	Recode Label	Recode Value	Code as Missing ?
1	Checkbox Checked	Yes	1	No
NA	Unanswered	No	0	No

# APPLY TO COLUMNS K:
COLUMNS <-  staffing_plans


# ALL RESPONSES NA IN GROUP:
# indication that the question
# was skipped by the respondent

row.id <- apply( survey_df[ COLUMNS ], 
                 MARGIN = 1, 
                 FUN = function(x) all(is.na(x)) ) 

# VALUES THAT NEED RECODING

RULES <- c(    
          
             "    1    =>>     1   ", 
             "   NA    =>>     0   "   )


rules <- parse_rules( RULES )          
pattern <- rules[[ "pattern" ]]
replace <- rules[[ "replace" ]]

# NA means "no"

na_to_zero <- function(x){
  x[ is.na(x) ] <- 0
  return(x)
}

survey_df[ COLUMNS ] <- 
  survey_df[ COLUMNS ] %>% 
  lapply( na_to_zero )

# if they do not respond "yes" 
# to at least one question then
# assume they skipped them all

zero_to_na <- function( x, rowid ){
  x[ rowid ] <- NA
  return(x)
}

survey_df[ COLUMNS ] <- 
  survey_df[ COLUMNS ] %>% 
  lapply( zero_to_na, row.id )

# MEMISC LABELS AND MISSING VALUE CODES 
values  <- c( "0", "1" )
labels  <- c( "No", "Yes" )
missing <- NA

# RECODE VARIABLES 
survey_df <- 
  survey_df %>% 
  recode_columns( k=COLUMNS, pattern, replace, values, labels, missing )

Example:

StaffingPlans_HireDifferent — ‘TYPE: boolean’

“Indicates whether the nonprofit has plans to hire new staff with different skills than existing staff in the next 12 months”

Storage mode:	character
Measurement:	nominal
Missing values:	NA

Values and labels			N	Percent
0		‘No’	329	65	.	5
1		‘Yes’	173	34	.	5

QUESTION TXT:

Does your organization have plans to change its staffing in the next 12 months? (select all that apply) - Hire new staff with different skills than existing staff

10.4 N/A Checkboxes

These questions are presented as a checkbox to the user to indicate that a question is not applicable. “Yes” here means “Yes, this question is not applicable”.

10.4.1 Staff Forecasts

We see three four cases:

• All answered
• Only answer for 2022 but don’t forecast 2023
• Answer “does not apply to me”
• Answer “does apply” but then don’t report numbers

We create an imputed version of 2023 numbers using 2022 numbers for missing cases (new variable x_2023 is saved as x_2023_imp for “imputed” ).

Replace “OK” in the not applicable field with “UNSURE” if they report no staff numbers. These will become NA (missing) when the dataset is saved.

    Staff_Fulltime_2022 Staff_Fulltime_2023 Staff_Fulltime_NA
142                 100                 110                 0
144                   1                   1                 0
145                 210                 225                 0
143                   1                  NA                 0
146                   0                  NA                 0
141                  NA                  NA                 1
150                  NA                  NA                 0

After imputation of 2023 responses (use 2022 values):

    Staff_Fulltime_2022 Staff_Fulltime_2023 Staff_Fulltime_NA
142                 100                 110                 0
144                   1                   1                 0
145                 210                 225                 0
143                   1                   1                 0
146                   0                   0                 0
141                  NA                  NA                 1
150                  NA                  NA                99

impute_staff <- function( df, v2022, v2023, vNA ){
  
  x.2022 <- df[[ v2022 ]]
  x.2023 <- df[[ v2023 ]]
  x.NA   <- df[[ vNA   ]]
  
  na.2022 <- is.na( x.2022 )
  na.2023 <- is.na( x.2023 )
  both.na <- na.2022 & na.2023
  
  # Not applicable is "unsure"
  x.NA[ is.na( x.NA ) ] <- 0
  x.NA[ x.NA == 0 & both.na ] <- 99
  
  # impute missing 2023 values with 2022 values 
  x.2023[ na.2023 ] <- x.2022[ na.2023 ]
  v2023i <- paste0( v2023, "_imp" )
  
  df[[ v2023i ]] <- x.2023
  df[[   vNA ]] <- x.NA
  
  return(df)
}


survey_df <- 
  survey_df %>% 
  impute_staff( v2022="Staff_Fulltime_2022", 
                v2023="Staff_Fulltime_2023", 
                vNA="Staff_Fulltime_NA" )

survey_df <- 
  survey_df %>% 
  impute_staff( v2022="Staff_Parttime_2022", 
                v2023="Staff_Parttime_2023", 
                vNA="Staff_Parttime_NA" )

survey_df <- 
  survey_df %>% 
  impute_staff( v2022="Staff_Boardmmbr_2022", 
                v2023="Staff_Boardmmbr_2023", 
                vNA="Staff_Boardmmbr_NA" )

survey_df <- 
  survey_df %>% 
  impute_staff( v2022="Staff_RegVlntr_2022", 
                v2023="Staff_RegVlntr_2023", 
                vNA="Staff_RegVlntr_NA" )

survey_df <- 
  survey_df %>% 
  impute_staff( v2022="Staff_EpsdVltnr_2022", 
                v2023="Staff_EpsdVltnr_2023", 
                vNA="Staff_EpsdVltnr_NA" )

survey_df <- 
  survey_df %>% 
  impute_staff( v2022="Staff_AmerVlntr_2022", 
                v2023="Staff_AmerVlntr_2023", 
                vNA="Staff_AmerVlntr_NA" )

survey_df <- 
  survey_df %>% 
  impute_staff( v2022="Staff_PdCnslt_2022", 
                v2023="Staff_PdCnslt_2023", 
                vNA="Staff_PdCnslt_NA" )

Original Value	Description	Recode Label	Recode Value	Code as Missing ?
1	Yes, this question is not applicable	Not Applicable	1	No
NA	No, this question applies to me	OK	1	No
All NA	N/A response conflicts	Unsure	99	Yes

# APPLY TO COLUMNS K:
COLUMNS <-  staff_qns_bool

# VALUES THAT NEED RECODING

RULES <- c(    
          
             "    1    =>>      1    ", 
             "    0    =>>      0    ", 
             "   99    =>>     99    "     )


rules <- parse_rules( RULES )          
pattern <- rules[[ "pattern" ]]
replace <- rules[[ "replace" ]]

# MEMISC LABELS AND MISSING VALUE CODES 
values  <- c( "0", "1", "99" )
labels  <- c( "OK", "N/A", "UNSURE" )
missing <- "UNSURE"

# RECODE VARIABLES 
survey_df <- 
  survey_df %>% 
  recode_columns( k=COLUMNS, pattern, replace, values, labels, missing )

Example:

Staff_Fulltime_NA — ‘TYPE: boolean’

“Number of full time staff not applicable”

Storage mode:	character
Measurement:	nominal
Missing values:	UNSURE

Values and labels			N	Percent
0		‘OK’	456	90	.	8
1		‘N/A’	40	8	.	0
99		‘UNSURE’	6	1	.	2

QUESTION TXT:

How many (paid and unpaid) people in your organization will have worke… - Check here if not applicable for your organization - Full-time paid staff (35 or more hours /week) - C

10.4.2 Reserve Questions

Original Value	Description	Recode Label	Recode Value	Code as Missing ?
1	Yes, this question is not applicable	N/A	1	No
NA	This question applies to me	0	OK	No
All NA	N/A response conflicts	Unsure	99	Yes

# APPLY TO COLUMNS K:
COLUMNS <-  reserve_qns_bool

survey_df[ COLUMNS ] <- 
  survey_df[ COLUMNS ] %>% 
  lapply( na_to_zero )

# did not respond to the reserve questions
no.resp <- is.na( survey_df$Reserves_Est )
survey_df$Reserves_NA_X[ no.resp ] <-  99
    

# VALUES THAT NEED RECODING

RULES <- c(    
          
             "    1    =>>     1   ", 
             "    0    =>>     0   "   )


rules <- parse_rules( RULES )          
pattern <- rules[[ "pattern" ]]
replace <- rules[[ "replace" ]]


# MEMISC LABELS AND MISSING VALUE CODES 
values  <- c( "0", "1", "99" )
labels  <- c( "OK", "N/A", "Did Not Answer" )
missing <- "99"

# RECODE VARIABLES 
survey_df <- 
  survey_df %>% 
  recode_columns( k=COLUMNS, pattern, replace, values, labels, missing )

Example:

Reserves_NA_X — ‘TYPE: boolean’

“Estimated total operating cash and unrestricted liquid assets in reserve is not applicable”

Storage mode:	character
Measurement:	nominal
Missing values:	99

Values and labels			N	Valid			Total
0		‘OK’	355	95	.	9	70	.	7
1		‘N/A’	15	4	.	1	3	.	0
99	M	‘Did Not Answer’	132				26	.	3

QUESTION TXT:

Approximately how much total operating cash and unrestricted liquid assets in reserve did you… - Select here if your organization does not have a reserve fund - Total operating cash and unrestricted liquid assets in reserve - C

10.4.3 People Served Questions

Original Value	Description	Recode Label	Recode Value	Code as Missing ?
1	Yes, this question is not applicable	N/A	1	No
NA	This question applies to me	0	OK	No
All NA	N/A response conflicts	Unsure	99	Yes

# APPLY TO COLUMNS K:
COLUMNS <-  people_served_qns_bool


# ALL RESPONSES NA IN GROUP:
# indication that the question
# was skipped by the respondent

people_qns <- c( "PplSrv_NumServed", "PplSrv_NumWait" )

row.id <- apply( survey_df[ people_qns ], 
                 MARGIN = 1, 
                 FUN = function(x){ all(is.na(x)) } ) 


# NA means "no"

na_to_zero <- function(x){
  x[ is.na(x) ] <- 0
  return(x)
}

survey_df[ COLUMNS ] <- 
  survey_df[ COLUMNS ] %>% 
  lapply( na_to_zero )

# if they do not respond "yes" 
# to at least one question then
# assume they skipped them all

zero_to_99 <- function( x, rowid ){
  x[ rowid ] <- 99
  return(x)
}

survey_df[ COLUMNS ] <- 
  survey_df[ COLUMNS ] %>% 
  lapply( zero_to_99, row.id )

    

# VALUES THAT NEED RECODING

RULES <- c(    "    1    =>>     1    ", 
               "    0    =>>     0    ",
               "   NA    =>>   _NA_   "      )


rules <- parse_rules( RULES )          
pattern <- rules[[ "pattern" ]]
replace <- rules[[ "replace" ]]


# MEMISC LABELS AND MISSING VALUE CODES 
values  <- c( "0", "1", "99", "_NA_" )
labels  <- c( "OK", "N/A", "No response", "No response" )
missing <- c( "99", "_NA_" )

# RECODE VARIABLES 
survey_df <- 
  survey_df %>% 
  recode_columns( k=COLUMNS, pattern, replace, values, labels, missing )

Example:

PplSrv_NumServed_NA_X — ‘TYPE: boolean’

“Number of people served not applicable”

Storage mode:	character
Measurement:	nominal
Missing values:	99, NA

Values and labels			N	Valid			Total
0		‘OK’	444	99	.	6	88	.	4
1		‘N/A’	2	0	.	4	0	.	4
99	M	‘No response’	56				11	.	2

QUESTION TXT:

Approximately how many people do you estimate your organization served and… - Check here if not applicable for your organization - Number of People Served - N/A

11 Multi-selection Inputs

These questions offer the user with multiple options to select one from. Since the options are ordered categories, they are coded on an ordinal scale.

11.1 Increase - Decrease Questions

There 2 questions that ask respondents to define changes via an increase or decrease relative to previous years. They are recoded on an ordinal scale.

11.2 Changes in Demand Questions

Original Value	Description	Recode Label	Recode Value	Code as Missing ?
Increase	Increase	Increase	2	No
Stay the same	Stay the same	Unchanged	1	No
Decrease	Decrease	Decrease	0	No
-99	Incomplete	No	99	Yes
NA	Unanswered	NA	NA	Yes

# APPLY TO COLUMNS K:
COLUMNS <-  demand_fct_qns

# VALUES THAT NEED RECODING

RULES <- c(    
          
             "   0    =>>     0   ",
             "   1    =>>     1   ", 
             "  98    =>>     2   "     )


rules <- parse_rules( RULES )          
pattern <- rules[[ "pattern" ]]
replace <- rules[[ "replace" ]]

# MEMISC LABELS AND MISSING VALUE CODES 
values  <- c( 2, 1, 0 )
labels  <- c( "Increase", "Unchanged", "Decrease" )
missing <- NA

# RECODE VARIABLES 
survey_df <- 
  survey_df %>% 
  recode_columns( k=COLUMNS, pattern, replace, values, labels, missing )

Example:

Dmnd_NxtYear — ‘TYPE: factor’

“Demand for programs or services next year”

Storage mode:	character
Measurement:	nominal
Missing values:	NA

Values and labels			N	Valid			Total
0		‘Decrease’	126	25	.	2	25	.	1
1		‘Unchanged’	7	1	.	4	1	.	4
2		‘Increase’	367	73	.	4	73	.	1
NA	M		2				0	.	4

QUESTION TXT:

In the next year, do you anticipate the demand for your programs to decrease, stay the same, or increase?

11.3 Changes in Fundraising Questions

Original Value	Description	Recode Label	Recode Value	Code as Missing ?
Increased significantly (by more than 10%)	Largest Increase	Increase Significantly	5	No
Increased moderately (by less than 10%)	Second Largest Increase	Increase Moderately	4	No
Stayed more or less the same	Third Largest Increase	Unchanged	3	No
Decreased moderately (by less than 10%)	Fourth Largest Increase	Decrease Moderately	2	No
Decreased significantly (by more than 10%)	Fifth Largest Increase	Decrease Significantly	1	No
Unsure	Unsure	Unsure	99	Missing
-99	Incomplete	Incomplete	98	Yes
N/A	Not Applicable	Not Applicable	97	Yes
NA	Unanswered	NA	NA	Yes

# APPLY TO COLUMNS K:
COLUMNS <-  fundraise_change_qns_fct

# VALUES THAT NEED RECODING

row.id <- apply( survey_df[ COLUMNS ], 
                 MARGIN = 1, 
                 FUN = function(x) all(is.na(x)) )

RULES <- c(    
          
     "   Increased significantly (by more than 10%)    =>>     5   ",
     "      Increased moderately (by less than 10%)    =>>     4   ",
     "                 Stayed more or less the same    =>>     3   ",
     "      Decreased moderately (by less than 10%)    =>>     2   ",
     "   Decreased significantly (by more than 10%)    =>>     1   ",
     "                                           98    =>>     0   ",
     "                                           99    =>>     X   "  )
 
rules <- parse_rules( RULES )          
pattern <- rules[[ "pattern" ]]
replace <- rules[[ "replace" ]]


all_na <- function( x, rowid ){
  x[ rowid ] <- NA
  return(x)
}

survey_df[ COLUMNS ] <- 
  survey_df[ COLUMNS ] %>% 
  lapply( all_na, row.id )


# MEMISC LABELS AND MISSING VALUE CODES 
values  <- c( 5, 4, 3, 2, 1, 0, "X"  )
labels  <- c( "Increase Significantly", "Increase Moderately", 
              "Unchanged", "Decrease Moderately", "Decrease Significantly", 
              "Unsure", "Not Applicable" )
missing <- c( "X" )

# RECODE VARIABLES 
survey_df <- 
  survey_df %>% 
  recode_columns( k=COLUMNS, pattern, replace, values, labels, missing )

Example:

FndRaise_Overall_Chng — ‘TYPE: factor’

“Change in overall donations”

Storage mode:	character
Measurement:	nominal
Missing values:	X

Values and labels			N	Valid			Total
0		‘Unsure’	10	2	.	2	2	.	0
1		‘Decrease Significantly’	62	13	.	4	12	.	4
2		‘Decrease Moderately’	41	8	.	9	8	.	2
3		‘Unchanged’	140	30	.	4	27	.	9
4		‘Increase Moderately’	136	29	.	5	27	.	1
5		‘Increase Significantly’	72	15	.	6	14	.	3
X	M	‘Not Applicable’	10				2	.	0
NA	M		31				6	.	2

QUESTION TXT:

To the best of your knowledge, has each of the following categories of donations to your org… - In 2022 compared to 2021 - Overall donations

12 Level of Importance Questions

There are 2 questions that ask respondents to rank the importance of volunteers and donors respectively. However, both sets of options’ are not identical. Hence, they are recoded to common values for reproducibility.

12.1 Volunteer Importance

Original Value	Description	Recode Label	Recode Value	Code as Missing ?
Essential - we depend entirely on volunteers to carry out our mission and goals	Maximum Importance	Essential	5	No
Very important - we depend on volunteers for a wide range of tasks, but not all	Second Most Important	Very Important	4	No
Somewhat important - we depend on volunteers for several key tasks	Third Most Important	Somewhat Important	3	No
Not very important - we depend on volunteers for only non-essential tasks	Fourth Most Important	Not Very Important	2	No
Not at all important - we could carry out our mission and goals without using volunteers	Fifth Most Important	Not At All Important	1	No
We do not use volunteers	Sixth Most Important	Not Used	0	No
-99	Incomplete	Incomplete	99	Yes
NA	Unanswered	NA	NA	Yes

# APPLY TO COLUMNS K:
COLUMNS <-  volimportance_qns_fct

# VALUES THAT NEED RECODING

RULES <- c(    

"           Essential - we depend entirely on volunteers to carry out our mission and goals    =>>     5   ",
"           Very important - we depend on volunteers for a wide range of tasks, but not all    =>>     4   ",
"                        Somewhat important - we depend on volunteers for several key tasks    =>>     3   ",
"                 Not very important - we depend on volunteers for only non-essential tasks    =>>     2   ",
"  Not at all important - we could carry out our mission and goals without using volunteers    =>>     1   ",
"                                                                  We do not use volunteers    =>>     0   ",
"                                                                                        99    =>>     0   "    )

rules <- parse_rules( RULES )          
pattern <- rules[[ "pattern" ]]
replace <- rules[[ "replace" ]]

# MEMISC LABELS AND MISSING VALUE CODES 
values  <- c( 5, 4, 3, 2, 1, 0 )
labels  <- c( "Essential", "Very Important", "Somewhat Important", 
              "Not Very Important", "Not At All Important", "Not Used" )
missing <- NA

# RECODE VARIABLES 
survey_df <- 
  survey_df %>% 
  recode_columns( k=COLUMNS, pattern, replace, values, labels, missing )

Example:

VolImportance — ‘TYPE: factor’

“Importance of volunteers to organization”

Storage mode:	character
Measurement:	nominal
Missing values:	NA

Values and labels			N	Valid			Total
0		‘Not Used’	31	6	.	2	6	.	2
1		‘Not At All Important’	111	22	.	2	22	.	1
2		‘Not Very Important’	134	26	.	7	26	.	7
3		‘Somewhat Important’	149	29	.	7	29	.	7
4		‘Very Important’	54	10	.	8	10	.	8
5		‘Essential’	22	4	.	4	4	.	4
NA	M		1				0	.	2

QUESTION TXT:

How important were volunteers – other than board members – to the work your organization conducted in 2021?

12.2 Donor Importance

Original Value	Description	Recode Label	Recode Value	Code as Missing ?
Essential, we depend entirely on individual donations to carry out our mission and goals	Maximum Importance	Essential	1	No
Very important, we depend on individual donations for a wide range of activities, but not all	Very Important	2	No
Important, we depend on individual donations for several key activities	Third Most Important	Somewhat Important	3	No
Not very important, we depend on individual donations for only non-essential activities	Fourth Most Important	Not Very Important	4	No
Not at all important, we could carry out our mission and goals without donations from individuals	Fifth Most Important	Not At All Important	5	No
We do not receive donations from individuals	Sixth Most Important	Not Used	6	No
-99	Incomplete	Incomplete	0	Yes
NA	Unanswered	NA	NA	Yes

# APPLY TO COLUMNS K:
COLUMNS <-  donimportance_qns_fct

# VALUES THAT NEED RECODING

RULES <- c(    

  "          Essential, we depend entirely on individual donations to carry out our mission and goals  =>>  1  ",
  "     Very important, we depend on individual donations for a wide range of activities, but not all  =>>  2  ",
  "                           Important, we depend on individual donations for several key activities  =>>  3  ",
  "           Not very important, we depend on individual donations for only non-essential activities  =>>  4  ",
  " Not at all important, we could carry out our mission and goals without donations from individuals  =>>  5  ",
  "                                                      We do not receive donations from individuals  =>>  6  ",
  "                                                                                               -99  =>>  0  "  )


rules <- parse_rules( RULES )          
pattern <- rules[[ "pattern" ]]
replace <- rules[[ "replace" ]]

# MEMISC LABELS AND MISSING VALUE CODES 
values  <- c( 1, 2, 3, 4, 5, 6  )
labels  <- c( "Essential", "Very Important", 
              "Somewhat Important", "Not Very Important", 
              "Not At All Important", "Not Used" )
missing <- NA

# RECODE VARIABLES 
survey_df <- 
  survey_df %>% 
  recode_columns( k=COLUMNS, pattern, replace, values, labels, missing )

Example:

DonImportance — ‘TYPE: factor’

“Importance of donors to organization”

Storage mode:	character
Measurement:	nominal
Missing values:	NA

Values and labels			N	Valid			Total
1		‘Essential’	125	25	.	0	24	.	9
2		‘Very Important’	218	43	.	5	43	.	4
3		‘Somewhat Important’	97	19	.	4	19	.	3
4		‘Not Very Important’	33	6	.	6	6	.	6
5		‘Not At All Important’	16	3	.	2	3	.	2
6		‘Not Used’	0	0	.	0	0	.	0
99		‘99’	12	2	.	4	2	.	4
NA	M		1				0	.	2

QUESTION TXT:

How important were donations from individuals (including direct donations and indirect donations through federated giving programs, United Way, and donor advised funds, or in-kind donations) to the work of your organization in 2021?

13 Frequency Questions

These questions ask respondents to rank the frequency at which they engage in an activity.

Original Value	Description	Recode Label	Recode Value	Code as Missing ?
More Often Than Not	Most Frequent	Most of the time	5	No
Frequently	Second Most Frequent	Frequently	4	No
Occasionally	Third Most Frequent	Occasionally	3	No
Rarely	Fourth Most Frequent	Rarely	2	No
Never	Fifth Most Frequent	Occasionally	1	No
-99	Incomplete	Incomplete	99	Yes
NA	Unanswered	NA	NA	Yes

# APPLY TO COLUMNS K:
COLUMNS <-  extaffairs_qns_fct

# VALUES THAT NEED RECODING

RULES <- c(    

     "    More Often Than Not    =>>     5   ",
     "             Frequently    =>>     4   ",
     "           Occasionally    =>>     3   ",
     "                 Rarely    =>>     2   ",
     "                  Never    =>>     1   "   )
     
rules <- parse_rules( RULES )          
pattern <- rules[[ "pattern" ]]
replace <- rules[[ "replace" ]]

# MEMISC LABELS AND MISSING VALUE CODES 
values  <- c( 5, 4, 3, 2, 1 )
labels  <- c( "Most of the time", "Frequently", 
              "Occasionally", "Rarely", "Never"  )
missing <- NA

# RECODE VARIABLES 
survey_df <- 
  survey_df %>% 
  recode_columns( k=COLUMNS, pattern, replace, values, labels, missing )

Example:

ExtAffairs_GenEd — ‘TYPE: factor’

“Frequency of activities aimed at educating the general public about a specific policy issue and/or the interests of certain groups”

Storage mode:	character
Measurement:	nominal
Missing values:	NA

Values and labels			N	Valid			Total
1		‘Never’	139	31	.	7	27	.	7
2		‘Rarely’	60	13	.	7	12	.	0
3		‘Occasionally’	119	27	.	1	23	.	7
4		‘Frequently’	91	20	.	7	18	.	1
5		‘Most of the time’	30	6	.	8	6	.	0
NA	M		63				12	.	5

QUESTION TXT:

During the last two years (2022-2023), how often did your organization conduct the following activities? (Including at the local, county, state, or federal level). - Educate the general public about a specific policy issue and/or the interests of certain groups

14 Integer Inputs

These questions accept an integer input from users to indicate the number of staff they have, people they served, or donors they have.

Original Value	Description	Recode Label	Recode Value	Code as Missing ?
Whole Number	Number of Staff, People or Donors	NA	Integer Value	No
N/A	Not Applicable	NA	-1	Yes
-99	Incomplete	NA	-2	Yes
NA	Unanswered	NA	NA	Yes

int_qns <- 
  c( staff_qns_int, 
     people_served_qns_int, 
     fundraise_donor_qns_int )

COLUMNS <- int_qns

survey_df[ COLUMNS ] <- 
  survey_df[ COLUMNS ] %>%
  lapply( recode_x, pattern=c("N/A","-99"), replace=c("Inf","Inf") )

survey_df[ COLUMNS ] <- 
  survey_df[ COLUMNS ] %>%
  lapply( as.numeric )

survey_df[ COLUMNS ] <- 
  survey_df[ COLUMNS ] %>%
  lapply( memisc::as.item, missing.values=Inf )

survey_df[ COLUMNS ] <- purrr::map( COLUMNS, add_q_details, survey_df )

Example:

Staff_Fulltime_2022 — ‘TYPE: integer’

“Number of full time staff”

Storage mode:	double
Measurement:	interval
Missing values:	Inf

Values and labels			N	Percent
NA	M		43	8	.	6

Min:	0	.	000
Max:	3800	.	000
Mean:	20	.	950
Std.Dev.:	179	.	420

QUESTION TXT:

How many (paid and unpaid) people in your organization will have worke… - 2022 - Full-time paid staff (35 or more hours /week) - (best estimate)

15 Numeric Inputs

These questions accept a numeric input from users to denote dollar values. For some variables, additional processing is done to remove commas and “$” signs.

Original Value	Description	Recode Label	Recode Value	Code as Missing ?
Dollar Amount	Dollar Amount	NA	Numeric Value	No
-99	Incomplete	NA	-1	Yes
NA	Unanswered	NA	NA	Yes

numeric_qns <- 
  c( majorgift_qn_num, 
     reserve_qns_num, 
     cares_qns_num, 
     finance_revenue_qns_num )

COLUMNS <- numeric_qns

survey_df[ COLUMNS ] <- 
  survey_df[ COLUMNS ] %>% 
  lapply( keep_numbers )

survey_df[ COLUMNS ] <- 
  survey_df[ COLUMNS ] %>%
  lapply( memisc::as.item, missing.values=Inf )

survey_df[ COLUMNS ] <- purrr::map( COLUMNS, add_q_details, survey_df )

Example:

FndRaise_MajGift_Amt — ‘TYPE: numeric’

“Donation amount considered a major gift”

Storage mode:	double
Measurement:	interval
Missing values:	Inf

Values and labels			N	Percent
NA	M		35	7	.	0

Min:	0	.	000
Max:	1000000	.	000
Mean:	6178	.	546
Std.Dev.:	47376	.	540

QUESTION TXT:

What is the smallest donation amount that your organization would consider to be a major gift? - $

16 Text Inputs

These questions allow the user to enter raw text as answers. All non-text values are converted to NAs.

text_qns <- 
  c( staff_qns_text, 
     finance_chng_qns_text, 
     finance_revenue_qns_text, 
     fundraise_qns_text, 
     leadership_chng_qns_text, 
     primary_cncrn_qn_text, 
     program_change_qns_txt, 
     race_gender_qns_text)

survey_df[ text_qns ] <- 
  survey_df[ text_qns ] %>%
  lapply( recode_x, pattern="-99", replace=NA )

survey_df[ text_qns ] <- 
  survey_df[ text_qns ] %>%
  lapply( memisc::as.item )

survey_df[ text_qns ] <- purrr::map( text_qns, add_q_details, survey_df )

codebook( survey_df[ text_qns[1] ] )

================================================================================

   Staff_Other_Text_2022 'TYPE: text'

   "Description of other staff"

--------------------------------------------------------------------------------

   Storage mode: character
   Measurement: nominal

   Min:                  "1099 artists"
   Max: "teaching artists & conductors"

   QUESTION TXT:
       How many (paid and unpaid) people in your organization will have
       worke... - 2022 - Other (please specify): - Text

Example:

Staff_Other_Text_2022 — ‘TYPE: text’

“Description of other staff”

Storage mode:	character
Measurement:	nominal

Min:	“1099 artists”
Max:	“teaching artists & conductors”

QUESTION TXT:

How many (paid and unpaid) people in your organization will have worke… - 2022 - Other (please specify): - Text