Section author: Rebecca Vederhus, Sebastian Jentschke

From SPSS to jamovi: Analysis of Covariance (ANCOVA)

This comparison shows how to conduct an analysis of covariance in SPSS and jamovi. The SPSS test follows the description in chapter 13.5.4 - 13.5.6 in Field (2017), especially figure 13.5 - 13.7 (excluding Options) and outputs 13.6 - 13.11 (excluding Bootstrap, as this is not [yet] an option in jamovi). It uses the data set Puppy Love.sav which can be downloaded from the web page accompanying the book.

SPSS

jamovi

In SPSS, you can run this test using: Analyze → General Linear Model → Univariate.

In jamovi you do this using: Analyses → ANOVA → ANCOVA.

SPSS_Menu_ANCOVA3

jamovi_Menu_ANCOVA3

In SPSS, move Happiness to the Dependent Variable box, Dose to the Fixed Factor(s) box, and Puppy_love into Covariate(s).

In jamovi, move Happiness to the Dependent Variable box, Dose to the Fixed Factors box and Puppy_love into Covariates.

SPSS_Input_ANCOVA3_1

jamovi_Input_ANCOVA3_1

Then, open the dialog box called Contrasts, and click the drop-down menu to select Simple. Change the Reference Category to First, and press Change.

Open the Contrasts window, and select simple from the drop-down menu.

SPSS_Input_ANCOVA3_2

jamovi_Input_ANCOVA3_2

Select EM Means from the sidebar. Move Dose to the box called Display Means for:, press Compare main effects and select Sidak in the drop-down list.

Then, move Dose to Term 1 in the Estimated Marginal Means window. Tick the box for Marginal means tables as shown in the picture below.

SPSS_Input_ANCOVA3_3

jamovi_Input_ANCOVA3_3

When comparing the output from SPSS and jamovi, the results are the same. However, SPSS provides a lot more, rather unnecessary output than jamovi. These outputs are not included here.

SPSS_Output_ANCOVA3_1

SPSS_Output_ANCOVA3_2

SPSS_Output_ANCOVA3_3

jamovi_Output_ANCOVA3

In SPSS, you can find the adjusted values of the group means in the Estimates table. By looking at the Mean you can find out if there are any changes in happiness levels if there is an increase in puppy exposure. The Tests of Between-Subjects Effects table shows the sum of squares for the Dose variable, and this tells us how many units of variance this factor account for. In the Contrast Results (K Matrix) table, level 2 (15 mins) is compared with level 1 (control) and then level 3 (30 mins) is compared with level 1 (control). Here, the group differences are indicated by standard error, a difference value and a p-value.

In jamovi, the adjusted values can be found in the table called Estimated Marginal Means – Dose. This table looks exactly the same as the equivalent table in SPSS. Sum of squares are found in the ANCOVA – Happiness table. These tables differ slightly in SPSS and in jamovi, as jamovi only gives results for Dose and Residuals. The contrast analysis in jamovi also does not provide all of the values that the SPSS analysis does. However, the most important information is included.

jamovi does not include the values for the Corrected Model, Intercept, Total and Corrected Total in its analysis of covariance. Apart from this, the ANCOVA table follows the same structure as in SPSS. The tables for Estimated Marginal Means are also identical in SPSS and jamovi. The results from the contrast analysis in jamovi appears much clearer than the one in SPSS. Here the results are presented horizontally, while the results in SPSS are presented vertically. In contrast to SPSS, jamovi only includes the following values: Estimate, SE, t and p.

The numerical values for these analyses are the same: SS*<sub>Dose</sub> = 25.19, *p < .05; SS*<sub>Puppy_love</sub> = 15.08, *p < .05; M*<sub>Control</sub> = 2.93, *M*<sub>15mins</sub> = 4.71, *M*<sub>30mins</sub> = 5.15; contrast 1, *p = 0.045; contrast 2, p = 0.010.

If you wish to replicate those analyses using syntax, you can use the commands below (in jamovi, just copy to code below to Rj). Alternatively, you can download the SPSS output files and the jamovi files with the analyses from below the syntax.

 
UNIANOVA Happiness BY Dose WITH Puppy_love
  /CONTRAST(Dose)=Simple(1)
  /METHOD=SSTYPE(3)
  /INTERCEPT=INCLUDE
  /EMMEANS=TABLES(Dose) WITH(Puppy_love=MEAN) COMPARE ADJ(SIDAK)
  /CRITERIA=ALPHA(0.05)
  /DESIGN=Puppy_love Dose.

jmv::ancova(
    formula = Happiness ~ Dose + Puppy_Love,
    data = data,
    contrasts = list(list(var = "Dose", type = "simple")),
    emMeans = ~ Dose,
    emmPlots = FALSE,
    emmPlotError = "none",
    emmTables = TRUE)

SPSS output file containing the analyses

jamovi file containing the analyses
References
Field, A. (2017). Discovering statistics using IBM SPSS statistics (5th ed.). SAGE Publications. https://edge.sagepub.com/field5e