LGBT Survey Analysis

Cover image credit: Wokandapix from Pixabay

Table of Contents

Note: I’ve hidden all the code blocks since they have taken so much place and have been somewhat distracting. I pushed the notebook to GitHub repo.

1. Project overview and objectives

The main purpose of this project is the visualization of survey results conducted in EU countries (and Croatia) among 93000 LGBT people (2012). I tried to estimate the overall score of “suitability” (in other words, how good is this county for LGBT community?) by assigning weights to answers and getting average scores for each of the question block. Then I look at some particular questions to explore how satisfied LGBT communiy is and what they think would improve their lives in the countries they live in.

1.1. The aim of the survey

The aim of the EU LGBT survey was to obtain robust and comparable data that would allow a better understanding of how lesbian, gay, bisexual and transgender (LGBT) people living in the European Union (EU) and Croatia experience the enjoyment of fundamental rights. The survey collected data from 93,079 people across the EU and Croatia through an anonymous online questionnaire, collecting the views, perceptions, opinions and experiences of persons aged 18 years or over, who self-identify as lesbian, gay, bisexual or transgender. The topics related to various fundamental rights issues with an emphasis on experienced discrimination, violence and harassment. The survey and all related activities covered the 27 current EU Member States as well as Croatia. FRA designed the questionnaire and finalised it in consultation with its Scientific Committee, relevant stakeholders and civil society organisations, as well as independent academics and national experts with expertise in the area of discrimination on grounds of sexual orientation and gender identity.

The survey asked a range of questions about LGBT people’s experiences including:

  • public perceptions and responses to homophobia and/or transphobia;
  • discrimination;
  • rights awareness;
  • safe environment;
  • violence and harassment;
  • the social context of being an LGBT person;
  • personal characteristics, including age and income group.

Taken from EU LGBT survey technical report. Methodology, online survey, questionnaire and sample

1.2. Data set overview

Data set consist of 5 .csv files that represent 5 blocks of questions.

The schema of all the tables is identical:

VariableNote/Example
CountryCodename of the country
subsetLesbian, Gay, Bisexual women, Bisexual men or Transgender
question_codeunique code ID for the question
question_labelfull question text
answeranswer given
percentage%
notes[0]: small sample size; [1]: NA due to small sample size; [2]: missing value
  • Total amount of countries that participated in the survey is 28
  • All answers are different (i.e, can be binary (Yes-No), numerical (1-10) or scale (Always-Often-Never))

Total number of questionsNumber of records with small sample size% of total records(0)Number of missing values due to the small sample size% of total records(1)Number of missing values% of total records(2)
Data set
Daily Life50.013447.039.51849.05.40.00.0
Rights Awareness10.0785.020.8130.03.40.00.0
Violence and Harassment47.025072.055.36897.015.20.00.0
Discrimination32.05782.036.71198.07.60.00.0

2. Choropleth map visualization of responses

This visualization allows to explore single question response by country. The dashboard was done using Microsoft Power BI. Original map visualization can be find here.

3. Country ‘suitable’ scores

In this section I am going to score each country by the survey answers to find out which county is “most suitable” for LGBT community. Each country will get a score in 4 blocks Daily Life, Discrimination, Violence and Harassment and Rights Awareness (I didn’t include Transgender Specific Questions here since the segment of people is transgenders only) and a final score.

3.1. Scoring methodology

First of all the ratio of Lesbians/Gays/Bisexuals/Transgenders are not equal amoung countries. In order to ‘normalize’ I am going to set the weight of each subset:

\begin{align*} \textrm{Weight}_{\text{subset}} = \frac{\text{# Subset for a Country}}{\text{# Total for a Country}} \end{align*}

Final Subset Weight values look like this:

After calculating the Subset Weight values I am going to get new value of Percent of responses for each subset by multiplying the original Percent value by the Subset Weight.

\begin{align*} \textrm{Percent}_{\textrm{weighted}} = \textrm{Percent} \times \textrm{Weight}_{\textrm{subset}} \end{align*}

After this I am adding a Response Weight which will show how ‘good’ the answer is. Let’s take a look at imaginary example for two qestions for Italy:

CountryQuestionAnswerPercent (Weighted)
ItalyIn your opinion, in the country where you live, how widespread is discrimination because a person is Transgender?Very widespread25
ItalyIn your opinion, in the country where you live, how widespread is discrimination because a person is Transgender?Fairly widespread15
ItalyIn your opinion, in the country where you live, how widespread is discrimination because a person is Transgender?Dont know10
ItalyIn your opinion, in the country where you live, how widespread is discrimination because a person is Transgender?Fairly rare30
ItalyIn your opinion, in the country where you live, how widespread is discrimination because a person is Transgender?Very rare20
ItalyHave you personally felt discriminated against or harassed because of being perceived as Gay?Yes30
ItalyHave you personally felt discriminated against or harassed because of being perceived as Gay?Don’t know20
ItalyHave you personally felt discriminated against or harassed because of being perceived as Gay?No50

First step is going to be adding a weight to each answer in range [-1, 1] with -1 being negative and 1 being positive. Looking at the first question In your opinion, in the country where you live, how widespread is discrimination because a person is Transgender? the answer Very rare is the best possible scenario among all the answer options while Very widespread is the worst. So I’m assigning weight -1 to Very widespread and 1 to Very rare. The rest weight of the answers are splited evenly (-0.5 to Fairly widespread and 0.5 to Fairly rare). For example if there is 6 answer options, the weights look like this [-1, -0.66, -0.33, 0.33, 0.66, 1]. Answer option Don't know gets np.NaN.

Note: before I thought that Don't know answer weight should be 0 but then I changed it to np.NaN so it doesn’t affect the total score since that answer is not really helpful. If you think it should be 0 I would love to hear your reasons.

Then I compute the Score by following formula:

\begin{align*} \textrm{Score} = \textrm{Weight}_{\textrm{response}} \times \frac{\textrm{Percent}_{\textrm{weighted}} }{100} \end{align*}

In that case Score can also be in the range [-1, 1] with -1 being negative and 1 being positive. The final Total Block Score for the country is just taking the average of all the scores.

\begin{align*} \textrm{Total Block Score} = \textrm{Average(Score)} \end{align*}

CountryQuestionAnswerPercent (Weighted)Response WeightScore
ItalyIn your opinion, in the country where you live, how widespread is discrimination because a person is Transgender?Very widespread25-1-0.25
ItalyIn your opinion, in the country where you live, how widespread is discrimination because a person is Transgender?Fairly widespread15-0.5-0.075
ItalyIn your opinion, in the country where you live, how widespread is discrimination because a person is Transgender?Don’t know10np.NaNnp.NaN
ItalyIn your opinion, in the country where you live, how widespread is discrimination because a person is Transgender?Fairly rare300.50.15
ItalyIn your opinion, in the country where you live, how widespread is discrimination because a person is Transgender?Very rare2010.2
ItalyHave you personally felt discriminated against or harassed because of being perceived as Gay?Yes30-1-0.3
ItalyHave you personally felt discriminated against or harassed because of being perceived as Gay?Don’t know20np.NaNnp.NaN
ItalyHave you personally felt discriminated against or harassed because of being perceived as Gay?No5010.2

So the `Total Block Score` for this block is going to be $\frac{-0.25 -.075 + 0.15 + 0.2 - 0.3 + 0.2}{6} = −0.0125$. After computing the scores for 4 blocks the `Total Score` is going to be the average of four `Total Block Scores`.

3.2. Daily Life

Let’s start with Daily Life questions block where subjects answered questions about day to day living as a lesbian, gay, bisexual or transgender person.

  • The first place goes to Netherlands🇳🇱 (which means that the responses about daily life for this country were more positive comparing to other countries).
  • The last place goes to Cyprus🇨🇾.

3.3. Right Awareness

  • The first place goes to Sweden🇸🇪 (which means that the people from the LGBT community are much more aware about their rights in that country comparing to other).
  • The last place goes to Greece🇬🇷.

3.4. Discrimination

  • The first place goes to Malta🇲🇹 (which means that the people from the LGBT feel less discriminated in that country comparing to other countries).
  • The last place goes to Romania🇷🇴.

3.5. Violence and Harassment

  • The first place goes to Finland🇫🇮 (which means that the people from the LGBT are beinge the subject of harassment or violation less often in that country comparing to other countries).
  • The last place goes to Estonia🇪🇪.

3.6. Overall rank

By taking the average of 4 scores we can rescale that values to get the final Total Rank.

So!

  • The absolute winners are Denmark🇩🇰, Netherlands🇳🇱, Sweden🇸🇪.
  • The absolute losers are Romania🇧🇬, Bulgaria🇧🇬, Cyprus🇨🇾.

Here is something to think about when you are considering a destination for travelling/relocation.

4. What the LGBT community says

After I got the Total Rank for each country I want to look at some particular responses to find out how does LGBT community respond to living in EU countries.

4.1. Do people fell satisfied in EU countries?

There was a question “All things considered, how satisfied would you say you are with your life these days?” in Daily Life questions block where subjects could pick a value from 0 to 10 (10 being the most satisfied) of how satisfied they feel. Using the same methodology I am going to find a score for this single question and compare it to the Total Rank from previous section.

\begin{align*} \textrm{Rank Diff} = \textrm{Satisfaction Rank} - \textrm{Total Rank} \end{align*}

In such way, - sign in Satisfaction Rank column means that LGBT community feel more satisfied in that county as I would guess from Total Rank value. + sign tells the opposite.

4.2. Are people being open about their orientation?

Next question “4 levels of being open about LGBT background” from Daiy Life questions block allow to see how open the LGBT community is in the country they live in. The possible answers are Never Open, Rarerly Open, Fairly Open, Very Open.

The countries in the plot are sorted by the Total Rank (the top countries have the highest rank, the bottom countries have the lowest rank). You can notice how the ‘openess ratio’ is correlated with country Total Rank - the higher the rank, the higher is the ratio of ‘open’ people.

  • Gay Men have the highest Very Open rate (23%) while Bisexual Men have the highest Never Open rate (75%).
  • In total, about 27% of people from LGBT community being open about their orientaion (Very Open + Fairly Open), especially in Netherlands🇳🇱 (15%).

4.3. What would allow to live more comfortable?

There were a series of questions “What would allow you to be more comfortable living as a LGB person?” with 8 different options that allow to explore what is missing in current situation in the country for the LGBT community to feel better.




  • High ratio of people (88%) agreed that Measures implemented at school to respect LGB people would improve the situation (especially in Italy🇮🇹 with 78%)
  • 16% of people feel satisfied with the The possibility to marry and/or register a partnership (top countries are: Netherlands🇳🇱, Belgium🇧🇪 and Portugal🇵🇹)
  • 9% of people don’t think that The possibility to foster / adopt children would change a lot.

5. Conclusions

So I estimated the country ranks of goodness for LGBT community, showed in what countries people are more open about their orientation and what do people think would make their life better. It’s just a small piece of insights that could be extracted from this survey so many more questions can be answered. You can also check the official report with survey analysis results.

Ruslan Klymentiev
Ruslan Klymentiev
Data-something-scientist

Trying to do useful things with the help of data and math

comments powered by Disqus

Related