# Job autonomy supervisor support and engagement in hybrid and onsite work

A research example illustrating the service. The items were written for this example.

## Study purpose

Examine the questionnaire structure and the associations of work resources with engagement, and compare engagement across hybrid and onsite work arrangements.

## Design and sample

A cross-sectional observational study of adult employees. Each row is one respondent measured once. The example has 360 records across two work arrangements and three tenure groups. Work arrangement was not randomly assigned.

## Research questions

- Do the 15 items empirically distinguish autonomy, supervisor support and engagement?
- How are autonomy and support associated with engagement separately and in a joint model?
- Does mean engagement differ between the two work arrangements?

## Hypotheses

- H1: Job autonomy is positively associated with work engagement.
- H2: Autonomy and supervisor support jointly predict engagement in a multiple linear regression compared with an intercept-only model.
- H3: Mean engagement is higher in the hybrid group than in the onsite group.

## Requested analyses

- Exploratory factor analysis of the 15 items, with suitability checks, factor-number assessment and oblique rotation; do not force three factors merely because the questionnaire has three sections
- Cronbach alpha for each scale and composite scoring after the preliminary measurement checks
- Sample, item and composite descriptives and a correlation matrix for the three scales
- H1: Pearson correlation of autonomy and engagement
- H2: Multiple linear regression of engagement on supervisor support and autonomy; examine residuals, error variance, collinearity and influential observations
- H3: An independent-group mean comparison of engagement, considering unequal variance and reporting an effect size
- Planned additional analysis: extend H2 with work arrangement and tenure groups to assess sensitivity to background adjustment; distinguish this from the three hypotheses
- Use two-sided tests at 0.05; treat the three hypotheses as a family and examine Holm-adjusted p values alongside the original p values

## Scale scoring

Each composite is the mean of its original five scale items. Higher scores indicate more of the construct. There are no reverse-scored items or missing values in this version. Do not change a scale solely to raise alpha; document and review an EFA result that challenges the intended structure before changing it.

## Limitations

The associations and group differences are observational and do not establish a causal effect of work arrangement. EFA and reliability in one sample are not external validation. The items are original demonstration items, not copies of validated questionnaires.

## Supplied files

`../data.csv` · `questionnaire.docx` · `codebook.md` · `../codebook.json`