[Combining data][2-way between-groups ANOVA]

#1
Dear smart people,

I am currently finishing my Master's in International Business Communication and I am wondering if I can combine data to increase the power of my stats.

For my research, I did the following:
-A 2x2 between-groups design, with as factors Fashion_Interest (high, low) and Text_Mode (Bilingual, Monolingual)
- Each of my respondents read one short blog entry that either contained anglicised words or were entirely written in Portuguese (so "bilingual" and "monolingual"). Because my supervisor asked me to, I had to make 2 sets of blog entries - one about a black outfit and one about a White outfit - the texts were diferente but about the same length, and the bilingual versions had the same amount of anglicisms. In the end, I had WhiteBilingual, WhiteMonolingual, BlackBilingual, BlackMonolingual, but each participant was only assigned one text - meaning: subjects exposed to entries completely written in Portuguese were not exposed to entries that contained anglicised words, and vice-versa, subjects who saw the White outfit didnn't se the black outfit and vice-versa).

For the purpose of my research, the color of the outfit doesn't matter. My analyses would become more powerful and less cumbersome if I just looked at the text mode and completely disregarded the "colour of the outfit". To have a good explanation for this decision, I thought about running independent t-tests to check if the variances and/or means are equal

What I did was select cases, so:
- Bilingual AND high interest in in fashion
- Bilingual AND low interest in fashion
- Monolingual AND high interest in fashion
- Monolingual AND low interest in fashion

For each of those selections, I ran a t-test using DressColour (black or White) as independent variable [AuthorCompetece as the dependente variable].

The results were:
- Bilingual AND high interest in in fashion: Equal variances assumed t(132) = 1.88, p = .062, Mwhite = 3.21, SDwhite =.65, Mblack = 3.00, SDblack =.83
- Bilingual AND low interest in fashion: Equal variances assumed t(142) = 3.12, p = .002, Mwhite = 3.10, SDwhite =.64, Mblack = 2.76, SDblack =.67
- Monolingual AND high interest in fashion: Equal variances not assumed t(146) = 1.44, p = .153, Mwhite = 3.12, SDwhite =.54, Mblack = 2.80, SDblack =.66
- Monolingual AND low interest in fashion: Equal variances assumed t(136) = 3.10, p = .002, Mwhite = 3.12, SDwhite =.54, Mblack = 2.80, SDblack =.66

If I am reading this correctly, for most cases, there are mean diferences but equal variances. For one case, there are no significant mean diferences.

Can I then simply combine the scores for the White and Black outfit? Does it make sense?

Thanks in advance :D