Developmental Relationships CFAs 2

Author

Landon Cox

Background

Measure Description

The Developmental Relationships (DR) 360 Search Institute (2021) is an 8-item measure designed to assess the quality of developmental relationships and the resources individuals receive through those relationships. Developmental relationships are close connections through which young people discover who they are, develop the ability to shape their lives, and learn how to interact with and contribute to the world around them.

Subscales:

Elements of a Developmental Relationship: assesses whether relational partners express care, challenge growth, provide support, share power, and expand possibilities.
Resources Acquired From Relationships: asseseses access to useful information, skill-building opportunities, and connections to other people who can support educational or career goals.

The DR 360 is intended to provide a snapshot of an individual’s relational web and can be used across different relational targets, such as peers, teachers, and mentors.

Items

Item Name	Item Text	Subscale
	These questions are about your most important mentor. This mentor…
`dr01`	…shows me that I matter to him/her.	Elements of a Developmental Relationship
`dr02`	…challenges me to be my best.	Elements of a Developmental Relationship
`dr03`	…listens to my ideas and takes them seriously.	Elements of a Developmental Relationship
`dr04`	…helps me accomplish tasks.	Elements of a Developmental Relationship
`dr05`	…introduces me to new experiences or opportunities.	Elements of a Developmental Relationship
`dr06`	…provides me with useful information for pursuing my education or career goals.	Resources Acquired from Relationships
`dr07`	…supports me in developing or strengthening the skills needed to pursue my education or career goals.	Resources Acquired from Relationships
`dr08`	…connects me with other people who help me pursue my education or career goals	Resources Acquired from Relationships

Data Preparation

Import Data

# Data import
dat <- read.csv("00_raw_data/data_help-seeking.csv") # Reads and stores original data

# Create data frame with only variables of interest
dat_vars <- dat |> 
  dplyr::select(starts_with("dr"))

Rename Variables

# Rename the columns
dat_vars <- 
  dat_vars |> 
  rename(
    dr01 = dr1_developmentalrelationships1_elements1,
    dr02 = dr2_developmentalrelationships2_elements2,
    dr03 = dr3_developmentalrelationships3_elements3,
    dr04 = dr4_developmentalrelationships4_elements4,
    dr05 = dr5_developmentalrelationships5_elements5,
    dr06 = dr6_developmentalrelationships6_resources.aquired1,
    dr07 = dr7_developmentalrelationships7_resources.acquired2,
    dr08 = dr8_developmentalrelationships8_resources.acquired3)

Rescale Variables

# Check data (Generates a stats summary for newly renamed variables)
varNames <- grep("dr", names(dat_vars), value = TRUE)
summary(dat_vars[,varNames])

# Loop the POM function over all columns of data
for (i in 1:length(varNames)) {
  dat_vars[,varNames[i]] <- POM_fixed(x = dat_vars[,varNames[i]],
                                      min_x = 0,
                                      max_x = 100)*5
}

CFAs

Model A

Model Specifications

Type: One-factor
Indicators: 8
Modifications: None

Factor Specification

## Developmental Relationships A (One Factor)
dev_relA <- '
# Define the latent factors.
dev_rel =~ dr01 +
           dr02 +
           dr03 +
           dr04 +
           dr05 +
           dr06 +
           dr07 +
           dr08
'

Measurement Model Results

Measure	Value
Model Fit Indices
χ²	77.037
df	20.000
p	0.000
CFI	0.918
TLI	0.886
RMSEA	0.161
SRMR	0.062

			Coeff	SE	95% CI	z	p
Factor Loadings
dev_rel	=~	dr01	0.861	0.028	0.806,0.916	30.721	0
dev_rel	=~	dr02	0.890	0.023	0.845,0.936	38.192	0
dev_rel	=~	dr03	0.881	0.025	0.832,0.93	35.258	0
dev_rel	=~	dr04	0.735	0.047	0.643,0.827	15.659	0
dev_rel	=~	dr05	0.754	0.044	0.668,0.841	17.116	0
dev_rel	=~	dr06	0.794	0.039	0.718,0.869	20.592	0
dev_rel	=~	dr07	0.902	0.022	0.86,0.944	41.860	0
dev_rel	=~	dr08	0.429	0.081	0.269,0.588	5.269	0

Model D

Model Specifications

Type: One-factor
Indicators: 7
Modifications: Remove dr08

Factor Specifications

## Developmental Relationships D (One Factor remove 'dr08')
dev_relD <- '
# Define the latent factors.
dev_rel =~ dr01 +
           dr02 +
           dr03 +
           dr04 +
           dr05 +
           dr06 +
           dr07
'

Measurement Model Results

Measure	Value
Model Fit Indices
χ²	23.531
df	14.000
p	0.052
CFI	0.985
TLI	0.977
RMSEA	0.079
SRMR	0.025

			Coeff	SE	95% CI	z	p
Factor Loadings
dev_rel	=~	dr01	0.868	0.027	0.816,0.921	32.476	0
dev_rel	=~	dr02	0.894	0.023	0.85,0.939	39.465	0
dev_rel	=~	dr03	0.882	0.025	0.833,0.931	35.635	0
dev_rel	=~	dr04	0.732	0.047	0.639,0.825	15.447	0
dev_rel	=~	dr05	0.746	0.045	0.657,0.834	16.481	0
dev_rel	=~	dr06	0.785	0.040	0.706,0.863	19.696	0
dev_rel	=~	dr07	0.901	0.022	0.859,0.944	41.453	0

Model F

Model Specifications

Type: One-factor
Indicators: 3
Modifications: Balanced parcels with dr08 removed

Parceling

## Developmental Relationships PB (Balanced Parceling, removed dr08)
# Create parcels using the Balancing Technique
dat_vars$drPB1 <- rowMeans(dat_vars[c("dr07", 
                                      "dr04")],
                                na.rm = TRUE)

dat_vars$drPB2 <- rowMeans(dat_vars[c("dr02",
                                      "dr05")],
                                na.rm = TRUE)

dat_vars$drPB3 <- rowMeans(dat_vars[c("dr03",
                                      "dr01",
                                      "dr06")],
                                na.rm = TRUE)

Factor Specification

dev_relF <- '
# Define the latent factors.
dev_rel =~ drPB1 +
           drPB2 +
           drPB3
'

Measurement Model Results

Measure	Value
Model Fit Indices
χ²	0
df	0
p	NA
CFI	1
TLI	1
RMSEA	0
SRMR	0

			Coeff	SE	95% CI	z	p
Factor Loadings
dev_rel	=~	drPB1	0.904	0.023	0.859,0.949	39.579	0
dev_rel	=~	drPB2	0.898	0.024	0.852,0.944	38.116	0
dev_rel	=~	drPB3	0.938	0.019	0.9,0.975	48.957	0

Fit Table

Figure 1 – Fit Table

Model	χ2	df	χ2∕df	p	CFI	TLI	RMSEA [90% CI]	SRMR	AIC	BIC
Model A	77.04	20	3.85	< .001	.92	.89	.16 [.12, .20]	.06	1,961.17	2,025.98
Model D	23.53	14	1.68	.052	.98	.98	.08 [.00, .13]	.02	1,591.27	1,647.98
Model F	0.00	0	Inf		1.0	1.0	.00 [.00, .00]	.00	607.72	632.03
Common guidelinesa	—	—	< 2 or 3	> .05	≥ .95	≥ .95	< .05 [.00, .08]	≤ .08	Smaller	Smaller
aBased on Schreiber (2017), Table 3.

Conclusion

A one factor and two factor model were tested and results from the two factor model do not support multiple dimensions due to only a small increase in overall fit over a one factor model, as well as a high covariance between both factors (0.939). Correlated residuals added to the one factor model in Developmental Relationships C, provided little improvement.

Item dr08 was removed in the Developmental Relationships D model due to distinctly poor loading in all models. This change improved fit to acceptable levels as shown in the fit table (Figure 1).

A balanced parceling technique was used for all 8 items as well as the 7-item model after removing dr08 in model Developmental Relationships PA and Developmental Relationsips PB, respectively. Again, the removal of dr08 resulted in improved parcel loadings.

Researchers intending to use the this measure are advised to use it as a unidimensional model and consider removing the 8th item, dr08.

References

Search Institute. 2021. Social Capital Assessment and Learning for Equity (SCALE) Measures: User Guide. Minneapolis, MN.