• No se han encontrado resultados

4. Resultados

4.6 Evaluación del Desempeño

4.6.1 Seguimiento, Medición, Análisis y Evaluación

The first part of the hypotheses test compares the framing of German climate change cover-age with the framing of Australian climate change covercover-age. This is why it uses the indices (see Table 21 to Table 26) and the results of the cluster analysis (see Table 29) presented in the first part of this chapter.

Hypothesis 1: Australia’s climate change coverage is dominated by frames of causes and effects as well as of action. Germany’s climate change coverage remains unspecific, just defining the problem.

For this first hypothesis, the result of the cluster analysis is used. For the German part of the hypothesis, the cluster “Disconnected problem definition” is relevant, because it represents unspecific climate change coverage and, thus, operationalises the German part of the hy-pothesis. The Australian part of the hypothesis is operationalised by, on the one hand, “Focus on consequences excluding effects on the economy” and “Economic consequences”, repre-senting frames of causes and effects. On the other hand, the clusters “Treatment recommen-dation – national”, “Treatment recommenrecommen-dation – international” and “Focus on treatment information” stand for frames of action.

To test hypothesis 1, the five clusters which represent the Australian climate change cover-age are summarised in a new cluster “Climate effects and action”. “Disconnected problem definition” remains, as it represents the German climate change coverage. The cluster “Scep-tics” is excluded from this test.

The contingency table (Table 30) shows that the percentage of “Disconnected problem def-inition” in Germany was higher (50.6 per cent) than in Australia (41.5 per cent). In Australia, the percentage of the new cluster “Climate Effects and Action” was higher (58.5 per cent) than in Germany (49.4 per cent).

The difference between the two countries is significant (p < .01). Hypothesis 1 is correct.

6 Analysis 167

Table 30: Dominant clusters in German and Australian climate change coverage

Cluster German

Media Australian

Media

n= 342 585

“Disconnected problem

definition” Number of articles (n)

Percentage within a country 173

50.6 243

41.5

“Climate effects and

ac-tion” Number of articles (n)

Percentage within a country 169

49.4 342

58.5

Chi2 value 7.1 Significance p < .01

Hypothesis 2: Politicians in Australia have a greater stake in the representation of the na-tional climate change discussion than those in Germany. They are more often mentioned in the country’s newspapers than German politicians in German newspapers.

Hypothesis 2 uses the index “Politicians” (see Table 26) to compare the appearance of poli-tical actors in the two countries. This is why polipoli-tical actors are given the data value “1”, while other actors are given value “0”. Table 31 confirms this hypothesis, showing that Aust-ralia’s politicians are the senders of 28.9 per cent of all climate change news, whereas in Germany only 11.7 per cent of all actors involved in climate change coverage are politicians.

This difference between the two countries is significant (p < .001).

Table 31: Appearance of politicians in German and Australian climate change coverage

Germany Australia

n= 375 637

“Politicians” Number of articles (n)

Percentage 44

11.7 184

28.9 No “Politicians”

mentioned Number of articles (n)

Percentage 331

88.3 453

71.1

Chi2 value 39.8 Significance p < .001

168 6 Analysis

Hypothesis 3: In Australia, the representation of climate-change-sceptic viewpoints is a nor-mal aspect of climate change coverage. German newspapers rarely give space to such scep-tical voices.

To test hypothesis 3, the six clusters which do not represent sceptical viewpoints on climate change are summarised in a new cluster “Non-sceptics”, representing the German climate change coverage in terms of hypothesis 3. “Sceptics” remains as it represents the Australian climate change coverage.

The contingency table (Table 32) shows that the percentage of articles presenting “Sceptical views” is similar in both countries. The small difference between the two countries is not significant (p = .49).

Hypothesis 3 is not substantiated.

Table 32: “Sceptical” viewpoints in German and Australian climate change coverage

Cluster German

Media Australian

Media

n= 375 633

“Sceptics” Number of articles (n)

Percentage 33

8.8 48

“Non-sceptics” Number of articles (n) 7.6

Percentage 342

91.2 585

92.4

Chi2 value 0.5 Significance p = .49

Hypothesis 4a: German newspapers represent a broad variety of perspectives on climate change. The diversity of frame sub-dimensions is high. In Australia, the variety of perspec-tives on climate change is low.

To test Hypothesis 4a, all indices introduced in Table 21 to Table 25 are used, but the index

“Problem definition”. It is excluded because “Problem definition”, at least, has to appear in each article and is, thus, not a sign for variety.

Then, all indices with a negative data value are consolidated in a new index group represent-ing non-human-induced climate change: “Denyrepresent-ing that humans cause climate change” (data value: -1) and “Natural causes” (data value: -1) from the index “Causes of climate change”

as well as “Effects on nature: changes in biodiversity – positive consequences” (data value:

-1) from the index “Effects on nature”, “Effects on the economy: economic benefits” (data

6 Analysis 169

value: -1) from the index “Effects on the economy”, “No treatment required: there is no problem” (data value: -2) from the index “Treatment recommendation: general”, “Something must be done, but should not influence economic growth” (data value: -1) and “No treatment required: the economy comes first” (data value: -2) from the index “Treatment recommen-dation: national focus” are summarised to this new index group. So, all frame sub dimensions which see climate change as natural or having positive consequences are now together in this new index group.

Second, all frame sub dimensions from the above-mentioned indices as well as from the indices “Effects on the weather”, “Effects on humans”, “Treatment recommendation: inter-national focus”, “Treatment recommendation: energy”, and “Treatment recommendation:

increased information” having a positive data value, so all frame sub dimensions which iden-tify climate change as a challenge, are summarise in another group.

Variety is found when one article includes more than one of these indices. So, when looking at hypothesis 4a, German newspapers are expected to represent more of these indices – so a broader variety – in their articles, whereas the expectation for the Australian climate change articles is that they mostly include only one index – so a low variety of perspectives.

An examination of the frequencies in both countries shows that 45.9 per cent of all articles have a low variety because they just include one of the indices. 14.0 per cent contain two indices, 4.9 per cent three and 2.0 per cent four indices, which represents a high variety. The new index group representing non-human-induced climate change and, thus, representing climate change as natural or having positive consequences, appears in 0.4 per cent of all articles. A total of 37.7 per cent of the articles just include a “Problem definition” and are, therefore, not part of this hypothesis test.

Comparing the variety of perspectives in Germany and Australia with the support of a t-test, the mean in Germany has a value of 0.83, while the mean in Australia has a value of 1.10.

Thus, the Australian climate change coverage has a broader variety than the climate change coverage in Germany. The difference between the two countries is significant (p < .001) (see Table 33).

170 6 Analysis

Table 33: Variety of climate change coverage in Germany and Australia

Germany Australia

n = 375 637

Mean (SD) Mean (SD) t-value / significance

Variety of climate

change coverage 0.83 (0.84) 1.08 (1.00) - 4.3 / p < .001 The scale ranges from 0 to 5.

Hypothesis 4b: The variety of actors in German newspapers is higher than the variety of actors in Australian newspapers.

This is why the indices of actors – who are “scientists”, “politicians”, “journalists”, “mem-bers of civil society”, “international actors”, and “economic actors” (see Table 26) – are used. A variety of actors is found when more than one of these indices, for example a scien-tist and a politician, are quoted in the same article.

An examination of the frequencies in both countries shows that 87.2 per cent of all articles only include one actor, so do not have a variety of actors. A variety of actors was found in 10.8 per cent of the articles with two actors, 1.1 per cent with three actors, and 0.1 per cent with four actors.

As Table 34 shows, a t-test identifies the fact that the variety of actors was higher in Aus-tralian than in German media. This difference is also significant (p < .001). So, the AusAus-tralian newspapers examined in this research do not only have a broader variety of perspectives on climate change, but also a broader variety of actors included in their climate change articles.

Table 34: Variety of actors involved in climate change coverage

Germany Australia

n = 375 637

Mean (SD) Mean (SD) t-value / significance

Variety of actors in

cli-mate change coverage 1.05 (0.28) 1.16 (0.44) -4.9 / p < .001 Each article could include a maximum of 4 actors. The scale ranges from 0 to 4.

6 Analysis 171

Hypothesis 5: German journalists use an informing writing style for climate change cover-age, while Australian journalists use a commenting one.

This hypothesis refers only to the actor group “Journalists”. Using the index “Journalists”, 380 articles are identified as having a journalist as an actor in German and Australian climate change coverage. A total of 167 of these articles (44.5 per cent) are found in German (n = 357) and 213 articles (33.4 per cent) in Australian newspapers (n = 637).

Looking at the form of expression of these articles – i. e. whether they are commenting or informing – a contingency table reveals that 52.1 per cent of all German articles with a jour-nalist as an actor were informing and 47.9 per cent were commenting. In Australia, 55.4 per cent were informing and 44.6 per cent were commenting (see Table 35). The difference be-tween the two countries is not significant (p = .52).

Table 35: Journalist use a commenting/informing writing style

Germany Australia

n= 167 213

Informing Number of articles (n)

Percentage 87

52.1 118

55.4 Commenting Number of articles (n)

Percentage 80

47.9 95

44.6

Chi2 value 0.4 Significance p = .52

Hypothesis 6: German as well as Australian climate change coverage asks national politi-cians to tackle climate change.

This hypothesis was deduced from the public’s attitude to climate change: The literature about the public’s attitude to climate change states that in Germany as well as in Australia people “say others are responsible” and name “politicians” as accountable for tackling cli-mate change. The public in both nations expects national politicians to act against clicli-mate change.

The frame element “Treatment recommendation” is used to test this hypothesis, because it contains frame sub-dimensions which involve asking national as well as international poli-ticians to act.

The frame sub-dimensions which ask national political actors to take action are “Rebuilding a natural environment”, “General reduction of greenhouse gases – national focus”, “Demand

172 6 Analysis

climate change policy – national focus”, “Demand more responsibility by the govern-ment/politicians”, “Reward climate-friendly behaviour”, “Economic aspects must be changed because of climate change (including greenwashing)”, “Something must be done, but should not influence economic growth”, and “No treatment required: the economy comes first”. These frame dimensions are given the positive data value “1”. The frame sub-dimensions which ask international political actors to tackle climate change include “Reduc-tion of greenhouse gases – interna“Reduc-tional” and “Demand climate change policy – interna-tional”. They get a negative data value “-1”.

To arrive at a useful variable, the international frame sub-dimensions are derived from the national ones: (“Rebuilding a natural environment” + “General reduction of greenhouse gases – national focus” + “Demand climate change policy – national focus” + “Demand more responsibility by the government/politicians” + “Reward climate-friendly behaviour”

+ “Economic aspects need to be changed because of climate change (including greenwash-ing)” + “Something must be done, but should not influence economic growth” + and “No treatment required: the economy comes first”) – (“Reduction of greenhouse gases – interna-tional” + “Demand climate change policy – internainterna-tional”). This new variable “Treatment recommendation – national versus international” is used for a t-test.

Table 36 illustrates that the German media has an international reference (Mean: -0.04), whereas the media in Australia concentrates on national action in terms of climate change (Mean: 0.17). The difference between Germany and Australia in terms of “Treatment rec-ommendation – national versus international” is significant (p < .001).

Table 36: “Treatment recommendation – national versus international” in Germany and Australia

Germany Australia

n = 357 637

Mean (SD) Mean (SD) t-value / significance

“Treatment

recommendation – na-tional versus

international”

-0.04 (0.41) 0.17 (0.59) -6.5 / p < .001

The scale ranges from -2 (Treatment recommendation: international frame sub dimensions) to 3 (Treatment recommendation: national frame sub dimensions) .

6 Analysis 173

This finding, nevertheless, shows that the public’s attitude to climate change – the German as well as the Australian people expect national politicians to take action against climate change – is not fully represented in the media. What was published in the newspapers in Germany was actually the opposite.

Hypothesis 7: German climate change coverage contains very few frame sub-dimensions related to the effects of climate change. In Australia, the media widely covers the effects of climate change.

Hypothesis 7 is deduced from the literature about the public’s attitudes and it is concerned with the impact climate change has on the people’s life.

To test this hypothesis, the frame element “Causal Interpretation (effects of climate change)”

is used. Bear in mind that this frame element was operationalized by four indices: 1. Effects on nature, 2. Effects on the weather, 3. Effects on humans, and 4. Effects on the economy.

Two indices have a negative data value: “Effects on nature: Changes in biodiversity – posi-tive consequences” and “Effects on the economy: economic benefits”; both have a data value of “-1”.

A t-test shows that only the difference regarding the index “Effects on the weather” is sig-nificant (p < .01).

Because this hypothesis does not examine the effects of climate change in detail, but rather ask generally whether the effects of climate change are mentioned, the four indices are, in addition, summarised in one index. Testing this new index “Effects of climate change, in general” with a t-test, we find a small but significant difference (p < .05) (see Table 37).

174 6 Analysis

Table 37: “Causal interpretation (effects of climate change)” in Germany and Australia

Germany Australia

n = 375 637

Mean (SD) Mean (SD) t-value /

significance

“Effects on nature” .13 (.42) .17 (.46) -1.2 /

p = .24

“Effects on the

weather” .07 (.29) .13 (.39) -2.9 /

p < .01

“Effects on

hu-mans” .08 (.29) .07 (.28) 0.5 /

p = .59

“Effects on the

economy” .07 (.35) .11 (.36) -1.5 /

p = .13

“Effects of climate

change, in general” .09 (.20) .12 (22) -2.2 /

p < .05

“Effects on nature”: scale ranges from -1 to 3; “Effects on the weather”: scale ranges from 0 to 2 ; “Effects on humans”: scale ranges from 0 to 2 ; “Effects on the economy”: scale ranges from -1 to 1 ; “Effects of climate change, in general”: scale ranges from -0.25 to 1.25

6.2.2. Comparisons between the two events

The second part of the hypotheses test compares the framing of the two events. So, it inves-tigates dominant frames/frame elements in the media coverage around the election and the environmental disaster, and it examines the depiction of the two events.

Sub research question 1: Does coverage during the environmental disaster emphasise na-tional activities, whereas coverage during the election emphasises internana-tional activities?

A t-test uses the new variable „Treatment recommendation – national versus international”

from Hypothesis 6 to find out whether coverage during the environmental disaster supports national activities and coverage during the election advocates international activities. This is why, on one the one hand, the time after the environmental disaster – the flooding – has happened is analysed. Because it is a national event, the assumption is that the newspaper coverage might discuss this environmental disaster in a national context. On the other hand, the time before the election is examined, i. e. the time during the election period when cli-mate change might be used for propaganda reasons.

Table 38 demonstrates that the assumption of sub research question 1 is right: the climate change coverage before the election demanded almost as much national as international ac-tion (Mean: 0.09), while the coverage after the environmental disaster focused on naac-tional action (Mean: 0.20). The finding is significant (p < .05).

6 Analysis 175

Table 38: “Treatment recommendation – national versus international” in the context of the election and the environmental disaster

Before Election After Environmental disaster

n = 287 168

Mean (SD) Mean (SD) t-value / significance

“Treatment recommen-dation – national ver-sus international”

0.09 (.50) 0.20 (.53) 2.4 / p < .05

Scale: -2 = ”Treatment recommendation international” / 3 = “Treatment recommendation national”

To provide more detailed results on when “causal interpretation (effects of climate change)”

(see Hypothesis 7) is part of the media discussion around the two events, the data set – again the time after the environmental disaster and the time before the election – is examined.

Sub research question 2: Does “causal interpretation (effects of climate change)”appear more often during the environmental disaster than during the political election?

The results of a t-test which concentrates on the two time periods – the time after the envi-ronmental disaster and the time before the political election – show that the four indices of

“Causal interpretation (effects of climate change)” as well as the summarised index “Effects of climate change, in general” reveal no major differences (see Table 39).

Table 39: “Causal interpretation (effects of climate change)” before the political election and after the environmental disaster

Before Election After Environmental disaster

n = 287 168

Mean (SD) Mean (SD) t-value / significance

“Effects on nature” 0.18 (.50) 0.14 (.42) - 0.9 / p = .35

“Effects on the

weather” 0.10 (.37) 0.16 (.40) 1.6 / p = .12

“Effects on humans” 0.06 (.26) 0.04 (.19) - 1.1 / p = .27

“Effects on the

econ-omy” 0.10 (.38) 0.07 (.34) - 0.9 / p = .39

“Effects on climate

change, in general” 0.36 (.57) 0.38 (.63) 0.4 / p = .71

Effects on nature: scale ranges from -1 to 3; . Effects on the weather: scale ranges from 0 to 2 ; Effects on humans: scale ranges from 0 to 2 ; Effects on the economy: scale ranges from -1 to 1 ; Effects of cli-mate change in general: scale ranges from -0.25 to 1.25

176 6 Analysis

Hypothesis 8: Whereas the newspaper coverage before the election demands action, the me-dia informs the public about the consequences of climate change after the time of the envi-ronmental disaster.

The assumption is that the newspaper coverage before the election is characterised by the clusters “Treatment recommendation – national, “Treatment recommendation – interna-tional” and “Focus on treatment information”, i. e. the three frames which stand for climate change action. The clusters “Focus on consequences excluding effects on the economy” and

“Economic consequences” illustrate the consequences of climate change and, thus, are ex-pected to be emphasised after the time of the environmental disaster.

Again the clusters are summarised to test hypothesis 8: “Treatment recommendation – na-tional, “Treatment recommendation – international” and “Focus on treatment information”

into the cluster “Climate action coverage”, “Focus on consequences excluding effects on the economy” and “Economic consequences” into the cluster “Climate consequences coverage”.

The contingency table shows that the percentage in “Climate action coverage” and “Climate consequences coverage” is very similar. “Climate action coverage” appears in 49.6 per cent of all articles during the time of the election and 45.6 per cent during the time of the envi-ronmental disaster. “Climate consequences coverage” is represented in 50.4 per cent of the climate change coverage around the election and in 54.4 per cent of all climate change cov-erage referring to the environmental disaster. The difference is not significant (p = .55). Hy-pothesis 8 is not substantiated (see Table 40).

Table 40: Comparison of “climate action coverage” and “climate consequences coverage” during the time of the election/environmental disaster

Cluster Before Election After Environmental

dis-aster

n= 137 90

“Climate ac-tion cover-age”

Number of articles (n)

Percentage 68

49.6 41

45.6

“Climate con-sequences coverage”

Number of articles (n)

Percentage 69

50.4 49

54.4

Chi2 value 0.4 Significance p = .55

6 Analysis 177

Sub research question 3: Is the newspaper coverage of climate change around the election commenting and is it informing during the time of the environmental disaster?

This sub research question is based on two assumptions: 1. That politicians increase their public relations activities during the election period, which is reflected in commenting arti-cles in the media, and 2. that an environmental disaster encourages journalists to provide facts and data to the public about climate change issues.

This is why the election period as well as the time after the environmental disaster are ex-amined with a contingency table.

The contingency table demonstrates that none of the events has a major effect on the form of expression. A total of 61.9 per cent of all articles published during the election period were informing and 62.7 per cent after the environmental disaster. Only 37.3 per cent of the articles published before the election and 38.1 per cent of the articles examined during the time after the environmental disaster were commenting. There is no significant difference between these results (p = .86) (see Table 41).

Table 41: Commenting/informing writing style before the election/after the environmental disaster

Cluster Before Election After Environmental

Disaster

Disaster